Allow freezing of Core.MethodTables

[fatteneder]

Implements the idea from #54138 (comment).

A follow-up PR would apply it to some relevant functions in Base.

[ararslan]

It could be nice to have analogous functionality for unsealing a method table as well.

[fatteneder]

i guess the analyzegc failure is due to me not being familiar with atomics ordering, will check tomorrow

[vtjnash]

SGTM, but would probably be good to hear from @JeffBezanson or @Keno or triage if they have thoughts about the API here? The gc-analyzer looks like you can fix it by adding assert(mt != NULL) there. I don't really know why it is triggering from this, but that should make it happy

There is also a core test that needs to be fixed (adding this field to the list of atomic fields)

[fatteneder]

The gc-analyzer looks like you can fix it by adding assert(mt != NULL) there. I don't really know why it is triggering from this, but that should make it happy

Thanks, that fixed it it locally. I could not make sense out of this either.

[nsajko]

This branch has conflicts that must be resolved

[vtjnash]

The feature may still be interesting, but the implementation details this relied on no longer exist, so there wouldn't be any way to rebase this